Why Katla is neither over nor dormant

There is a suggestion at Ice News that the volcano at Katla in Iceland is, and remains dormant. There is a similar article at the Volcanism blog noting that all is quiet. There are others including the Iceland Volcano and Earthquake blog and those who comment there, who believe, as I do, that the current events continue to presage a relatively strong eruption.

Let me step through some of the reasons for this, which involves me first going back to explaining the basic geology that is happening in Iceland, and then a little on the events that occur when rocks break, and what I conjecture then occurs post eruption, prior to the next time that one comes around. First I should add the caveat that one of the first things that one learns as a rock mechanic is that most rocks are different, i.e. there are very few cases where happened at one site is exactly repeated somewhere else, or the same the second time around, and rock properties change very rapidly over short distances, so that what I am about to write has a lot of generality to it, rather than being specific to this site. And I ask my colleagues to bear with me as I write what I recognize is a very simplified version of what happens, since I don't want to write a book chapter.

There is a certain amount of repetition in this post, of information that I have written over the past fifteen months, and for this I crave your indulgence. Nevertheless, looking at the evolution of the earthquakes around Katla since January of this year, it is clear that a magma passage has been created, which is still being developed, and will likely continue in that mode until an eruption occurs, with the likelihood that the longer it is now delayed, the larger that it will be, as I will explain below.

The volcanic complex at Eyjafjallajokull/Mrydalsjokull in Iceland is located where the Earth’s plates are moving relative to one another, and as they do so and create passages, magma from deeper within the Earth can make its way to the surface, giving the eruptions that occur from time to time. These occur through the volcanoes that line the two rifts that join in the middle of the island, Eyjafjallajokull and Katla, which lies under the Myrdalsjokull glacier, lying along the Eastern rift and on the Eurasian plate.

Illustration of the rifts through Iceland, and the location of the volcanoes (London Times)

Let me try a very simple analogy to try and explain what happens. Think of a sheet of plywood, floating in a bowl of hot glue. As I pull the sheet apart it will start to tear at some point within the sheet. The tear won’t be straight or initially continuous, and as it happens the tear will open a path for the underlying glue to flow up through the split to the surface, where it spills out, and then cools and hardens, re-attaching the two sides together and sealing over the underlying, still molten glue. The movement of the wood as it tears, lowers the forces (stresses) held within the sheet, and so, initially the wood becomes stable again. But as the sheet continues to be pulled apart, the stresses start to build again and that initial tear is now a weaker spot in the surface, and so when the pull again becomes too strong, the sheet will fail again in about the same place (the weakest link failing), opening a tear, glue will rise, sealing and reattaching the two surfaces, and the process repeats. And so, over time these points where the weakest links are located become, in the real equivalent, where volcanoes erupt at intervals. This is only partially valid as an analogy, since rock and wood fail in different ways, as I will explain, but it hopefully allows you to see a volcanic eruption as a part of a sequence of events, rather than just looking at the eruption as a single point in time.

So, if one begins with that gross simplification of events, consider now what has been happening in the region around Katla. When one applies a load to a rock sample it will start to fail normally somewhere at about 50% of the load at which it will totally fracture into pieces. The onset of that failure is detected because the cracks that are naturally found within the rock start to grow, and as they do they make sounds, and release small amounts of energy. At first the cracks that grow are somewhat randomly distributed through the sample, as any flaws that are favorably aligned to the growing stresses in the rock fail. Within a small 2-inch diameter core normally tested in the lab, this initial fracturing can only be detected with instruments and it is only when the rock gets to perhaps 80% of final load that cracking can be audibly detected. However, in the larger scale which is occurring in Iceland, these initial failure cracks are large enough that they release enough energy to be considered as small earthquakes when they occur. Initially, as the movement of the plates acts on the rock at their edges, in the same way as with with the rock core tested in the lab, these local failures and quakes are distributed throughout the region that is being stressed. And if one looks at the pattern of quakes along the rift region near Katla last February, for example, then one saw that the quakes were scattered around the region.

The scattered distribution of earthquakes in the Katla region of Iceland in February (source Icelandic Met Office)

At this time the quakes are occurring in the same way as we see the initial cracks starting to grow when we test a rock sample in the lab. There are some pre-existing weaknesses in the rock that are set so that as the load on the rock grows (the two plates are moving relatively steadily one to the other) that fail, and at these loads the distribution tends to be more random (although it also includes the weakness planes created during the last eruption).

As the cracks grow they relieve the surrounding rock of some stress, so that the load redistributes, but as the load continues to grow, so the longer weaker planes created in previous eruptions come to be more the weakest links in the surrounding rock structure, and the crack development (being simplistic) which generates the quakes begins to concentrate in location on the pre-weakened and damaged zones created in the run-up to previous earthquakes, the weakest links. So that if we look at the quake locations in April and May, one can see the regions where the quakes are occurring beginning to shrink down and focus around what turns out to be three regions.

Quakes around the region in April

Quakes in the same area in May

By last month, June, it was clear that the quakes were concentrating more on these regions and so I color coded the quakes for periods of ten days at a time.

Volcanic activity at the Katla volcano site in Iceland in June 2011. Red were in the first ten days, green in the second, and blue in the final eleven days of the month. Katla lies under the glacier Myrdalsjokull. (info from Icelandic Met Office)

The zones of concentration are in the caldera region of Katla (the bowl in the middle of the volcano), about half way between Eyjafjallajokull and Katla, and an odd grouping down near the sea.

So now we come to the first ten days of this month, and the flood of melt water from Katla. And this is where I want to explain why I am not convinced by the expert opinions suggesting there is no problem.

If one looks at the quake activity in the 24 hours before the flood, (which occurred at about 4 am Saturday the 9th) there was a flurry of small quakes in the caldera of the volcano.

Katla quakes in the last 24 hours (Icelandic Met Office )

The tear in the rock is focusing on the weak zone and beginning to open the crack up. Bear in mind that the magma under the surface rock (like the hot glue I mentioned in the beginning) can only get out through a weakened zone in the rock, and this has been developing over the past few days as the quakes show the development of a broadly damaged done under the caldera (as well as the one half-way to Eyjafjallajokull, though that is not as pronounced). Then, as Jon has noted at the Iceland Volcano and Earthquake blog, harmonics develop within the ground as the magma starts to move and the cycle are moving toward an eruption.

So now the question is, was the relatively small melt of the glacier ice that flooded the road indicative that the magma had flowed through the broken zone, and then cooled to seal off the cracks, even as the cracks had locally reduced the stress in the surrounding rock so that the pressure was relieved and the ground would become stable for another few decades.

I believe we already have the answer to that. If we look at the pattern of quakes after the event, looking at the Icelandic Met Office site as I write this at 9:30 pm Sunday, the pattern of quakes continues in the same places as it did prior to the flood.

Quakes in the last 24-hours (Icelandic Met Office)

For the fissures to be sealed and the stress distributed I expect that there will now have to be an eruption of magma to permeate the fractured zones of rock so that they no longer become susceptible to continued growth, as the overall stress on the rock is relieved. (The flow of hot glue that I mentioned at the top of the post).

As long as the quakes remain focused in the regions that they now are, those regions will continue to act as zones of stress concentration and thus further failure and weakening, until they are sealed by a larger eruption. And I expect that this will come sooner rather than later.

And, since the stress continues to build, vide the quakes continuing to happen, and the ground becomes more fractured as evidenced by those quakes, then the larger the passage will be through the ground and the greater the eruption.